Automated fact-checking of climate claims with large language models.

Area of Science:

• Environmental Science
• Artificial Intelligence
• Information Science

Background:

• The digital age presents challenges in distinguishing true from false climate information.
• Misinformation impacts public perception and climate policy.
• Automated fact-checking is crucial for validating climate claims.

Purpose of the Study:

• To develop an advanced automated fact-checking system for climate change information.
• To leverage Large Language Models (LLMs) for improved accuracy in fact-checking.
• To introduce Climinator (CLImate Mediator for INformed Analysis and Transparent Objective Reasoning) for real-time misinformation identification.

Main Methods:

• Utilizing state-of-the-art Large Language Models (LLMs).
• Integrating authoritative and up-to-date data sources.
• Implementing a novel debating framework for context-aware analysis and incorporating multiple scientific viewpoints.

Main Results:

• Significantly improved performance in automated fact-checking of climate claims.
• Climinator provides trustworthy and context-aware analysis.
• Real-time identification of climate misinformation.

Conclusions:

• Climinator enhances the reliability of automated fact-checking for climate information.
• The AI-driven debating framework facilitates informed dialogue on climate change.
• Highlights the potential of AI in environmental policy and data-driven discussions.